Bayesian Learning and Montecarlo Simulation - US GDP & Inflation

Grade: 30L

Project Overview

This project examines two crucial economic indicators of the United States: the Gross Domestic Product (GDP) and the Consumer Price Index for All Urban Consumers (CPIAUCSL) from 1948 to 2021. Both metrics are reported quarterly and have been seasonally adjusted. The primary goal is to analyze their percentage changes over time using various models.

Objectives

1. Fit each time series independently using AR, MA, ARMA, and GARCH models.
2. Fit the two time series jointly using a VAR model.
3. Use these models for in-sample and out-of-sample predictions.
4. Compare different models using DIC and WAIC criteria.

Methodology

All models were implemented using JAGS with the following specifications:

• 3 Chains
• Total of 10,000 Iterations
• 1,000 Burn-in Iterations

Data Handling

The data fed into JAGS consisted of only the first 90% of each time series. The remaining 10% was used to assess the out-of-sample predictions generated by the models.

Repository Structure

• 'assignment/' - Contains the PDF with the possible projects and the project delivery details.
• 'dataset/' - Contains the data of the two time series.
• 'output/' - Contains the delivered report and the presentation prepared for presenting the results.
• 'report/' - LaTeX code for a detailed report of the work performed.
• 'src/' - Source code for data analysis, model fitting, and evaluation.

Prerequisites

• JAGS
• R